I don’t quite understand what you mean. Does your classification model support both people and faces as input? you can do like this
pgie(person) --> sgie --> pgie(face) --> sgie
If you want to detect people first and then detect faces from people, and the classification model uses faces as input
You can refer to back-to-back-detectors then use face as classifier input
There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks